July 25, 2018
Four members of the Borsuk lab and Dr. Mark Borsuk attended the International Congress on Environmental Modelling and Software. All four members that attended, Ryan Calder, Kim Bourne, Jon Holt and Chris Krapu, gave oral presentations on current projects of theirs at the conference.
Kim presented her ongoing work exploring the contamination of both mercury and organics in fish tissues. Recently, there have been questions of whether current methods of determining fish advisories in the United States are meeting our maximum acceptable risk standards. Many of these advisories do not consider multiple contaminants, leading to a possible underestimation of risk. Kim’s project aims to explore the occurrence of co-contamination of mercury and organic contaminants using a Bayesian method called Generalized Joint Attribute Modeling. Kim has collaborated with many scientists to complete this project including Dr. Celia Chen of Dartmouth, Amanda Curtis of the University of Illinois – Urbana Champaign, and Jim Clark of Duke.
Chris also presented novel methods for Bayesian estimation. For many models, he needs to estimate parameters of a mechanistic model using statistical methods. This presents many challenges when using Bayesian estimation due to parameters often being high-dimensional or correlated. However, using a Hamiltonian Monte Carlo (HMC) inference method increases the efficiency of Bayesian estimation of such parameters. Chris used a hydrology model in Theano as a proof of concept for this methodology and found that HMC methods successfully estimated coherent parameters for a range of hydrology models.
Jon used the opportunity to present his work on family forest owner (FFO) decision making in New England. With impending risks to our forests from insects and other blights, it is important to understand how the managers of our landscape will react to these threats. Jon has used survey and census data to define agent functional types (AFTs) of those that will always cut down the forest as a precautionary measure, those who would never cut down their forest, and those that will respond to threats depending on the severity and immediacy of the threat. The actions of these AFTs have a large influence on ecosystem response of our forest regions in the Northeast as climate change progresses and tree-killing insects progress northward.
Ryan’s presentation also addressed risks and adaptations to climate change but focused on issues of sea level rise. By 2100, it is expected that half a million more people in California will be in the 100-year flood zone. The restoration of wetlands could minimize the damage due to this change. This work that Ryan and colleague Congjie Shi have completed in collaboration with the Nicholas Institute for Environmental Policy Solutions aims to predict the economic and ecological benefits of wetland restoration in the wake of sea level rise using an ecosystem services model. A case study was completed for Gallinas Creek in Marin County, California in which the benefits of upland stormwater retention, tidal storm surge dissipation, enhanced ecological productivity and recreational value in the community from wetlands were valued. The benefits of wetland restoration are variable based on location, but this model can help guide decision making regarding wetland restoration more broadly.